Count parameter for hcluster

It would be real handy if you could add a count_table parameter to hcluster(). Cheers,

Do you actually find hcluster useful? We’ve kind of given up on it since it explodes with big datasets when using the preferred average neighbor approach.

I must admit I have not tried it. I was looking for an alternative clustering algorithm that would allow me to cluster the ca 160000 unique reads I get for my experiment following your MiSeq SOP without using >100 GB of RAM. Hcluster() sounded promising. I guess I’ll just remove any remaining singletons after preclustering - that more than solves the RAM problems I encountered. Simon